CONTROL POINTS

New comment
 

I would like insert the control point model in my expert advisor.
Have you the algorithm to determinate the control point, the explanation of metaquote is not clear, perhaps it's a secret algorithm :)

Thanks
Fabien_math

 
The market does not move according to algorithm of generation of Сontrol Points. Therefore I do not understand what for to you it it is necessary to know. I personally did not study this question.
 
Rosh:
The market does not move according to algorithm of generation of Сontrol Points. Therefore I do not understand what for to you it it is necessary to know. I personally did not study this question.

dear Rosh, If I use the signal of my expert with a filtration with Control Point the result is more better, I thing this is due by an effect of attenuation of the noise if you filter the signal of your expert with the Control Points.

The majority of the model give bad results in trading real but with my great surprise I noted that if you filters the result of your model by Control Point then the result
is much better.

Then I would like introduce directly the algorithme of control Point in my expert code.

Sometimes even I noted that the association model/Control Point anticipated the result future of a few minutes or a few hours.


I thing it is excellent way to explore ...

If you have this control point algorithm I will be able to make you a demonstration afterwards.

thanks
Fabien_math
 
fabien_math:
Rosh:
The market does not move according to algorithm of generation of Сontrol Points. Therefore I do not understand what for to you it it is necessary to know. I personally did not study this question.

dear Rosh, If I use the signal of my expert with a filtration with Control Point the result is more better, I thing this is due by an effect of attenuation of the noise if you filter the signal of your expert with the Control Points.

The majority of the model give bad results in trading real but with my great surprise I noted that if you filters the result of your model by Control Point then the result
is much better.

Then I would like introduce directly the algorithme of control Point in my expert code.

Sometimes even I noted that the association model/Control Point anticipated the result future of a few minutes or a few hours.


I thing it is excellent way to explore ...

If you have this control point algorithm I will be able to make you a demonstration afterwards.

thanks
Fabien_math

Hey Fabien_math,

I've asked the same question see here 'open and close trade at control points'
I now understand that control point testing is nothing, because the actual price movement within a bar in real trading is not counted and it in fact cannot be modelled. For example, if you test H4, where many Martingale EAs work on this timeframe, you may yield exceptionally good results testing on control points with the price (high, low, open, close) at each control it modelled, especially if u test timeframes larger than M5 , and you thought your EA work. BUT, in reality, you never know if price move from open to high first or to low first, and in larger timeframes, that could make a huge difference, say a winning sell in the test (High to Low) may yield a losing sell because in reality price moves from Low to High. . .
So my suggestion is: forget about control points, test the every tick mode. And you shall be very happy if you can get half good in real trading as the results you get in the test. Just my two cents worth:)
New comment